IoT-Based Smart Jacket with Machine Learning for Real-Time Health Data Analysis

Overview

This project develops a wearable smart jacket that monitors health metrics (heart rate, body temperature, activity levels) in real-time using IoT sensors. Machine learning (ML) analyzes data to detect anomalies (e.g., irregular heartbeat, fever, fatigue), sending alerts to a mobile app or cloud platform. Designed for Indian students new to IoT, it’s affordable, practical for healthcare in rural/urban India, and ideal for competitions like TechNova 2025.

Motivation

India’s healthcare access is limited, especially in rural areas. This smart jacket offers a low-cost solution for continuous health monitoring, benefiting the elderly, athletes, or patients with chronic conditions. It detects health risks early (e.g., high heart rate, heat stress) and supports remote monitoring via a phone app, aligning with Digital India’s affordable tech goals. For students, it’s a hands-on way to learn IoT, sensors, and ML.

Features

• Real-Time Monitoring: Tracks heart rate, temperature, and activity using wearable sensors.
• Anomaly Detection: ML identifies irregular patterns (e.g., heart rate > 100 bpm, temp > 38°C).
• IoT Connectivity: Sends data to a mobile app (Blynk) or cloud (ThingSpeak) for visualization/alerts.
• Wearable Design: Sensors embedded in a comfortable jacket using conductive thread.

Components Needed

Below are the components, with beginner-friendly explanations. All are available from Amazon.in, Robu.in, or local markets (e.g., Lamington Road, Mumbai; SP Road, Bangalore).

• Microcontroller:
  • ESP32: Small chip with Wi-Fi/Bluetooth for IoT. Easy to program with Arduino IDE. Cost: ₹400-500.
  • Alternative: Arduino Nano (₹300, needs extra Wi-Fi module) or Raspberry Pi Zero (₹1,200, more power).
• Sensors:
  • MAX30102 Pulse Oximeter: Measures heart rate (HR) and blood oxygen (SpO2) via light on wrist/chest. Uses I2C connection. Cost: ₹300-500.
  • DS18B20 Temperature Sensor: Measures body temperature (sewn near armpit). Waterproof version for fabric. Cost: ₹100-150.
  • MPU-6050 Accelerometer: Tracks steps or posture (fall detection). 6-axis sensor. Cost: ₹200-300.
  • Optional: AD8232 ECG (heart rhythm, ₹500-800) or DHT11 (sweat, ₹100).
• IoT Connectivity:
  • ESP32’s built-in Wi-Fi or ESP8266 module (₹250-300). Use free platforms: ThingSpeak (graphs) or Blynk (app).
• Power Supply:
  • 3.7V Li-ion battery (2000mAh, ₹300) with TP4056 charger (₹100). Lasts ~8-12 hours. Beginner Tip: Test with USB power bank (₹300).
• Display/Output:
  • 0.96" OLED (SSD1306, ₹300-400) for on-jacket HR display. Buzzer (₹50) or LED (₹20) for alerts.
• Jacket & Integration:
  • Cotton jacket (₹400-600, local store). Conductive thread (₹100-200) or sewable boards (e.g., Adafruit Flora, ₹500).
• Miscellaneous:
  • Breadboard (₹100), jumper wires (₹50), resistors/capacitors (₹50). Enclosure/pouch (₹100) for protection.
• Total: ~10-14 components. Prototype on breadboard first, then sew into jacket.

Sourcing Tips: Check college labs for free components. Bulk buy from local markets saves 10-20%.

Budget

• Basic Prototype: ₹2,000-3,000 (~$24-36). ESP32, basic sensors (HR, temp, accelerometer), battery, jacket, breadboard.
• Mid-Range Prototype: ₹3,500-5,000 (~$42-60). Adds OLED, optional ECG, cloud/app setup.
• High-End Prototype: ₹6,000-8,000 (~$72-96). Raspberry Pi Zero, advanced sensors, custom app, sewn design.
• Breakdown:
  • Sensors: 40% (₹600-1,200).
  • Microcontroller/IoT: 20% (₹400-800).
  • Power/Display: 20% (₹500-800).
  • Jacket/Misc: 20% (₹500-1,000).
• Beginner Tips: Start basic (₹2,000). Use free cloud platforms (ThingSpeak/Blynk). Borrow tools (soldering iron, ~₹500) from labs.

Difficulty

• Level: Intermediate (6/10). Suitable for 2nd-3rd year ECE/CS students new to IoT. Takes 2-4 weeks.
• Breakdown:
  • Hardware: Moderate. Connect sensors to ESP32 via breadboard (easy with tutorials). Soldering/sewing into jacket needs practice (test on spare fabric). Keep weight ~200g for comfort.
  • Programming: Moderate. Arduino IDE for ESP32 (sensor reading, Wi-Fi). Blynk app setup is drag-and-drop. Python for cloud ML if needed.
  • IoT: Easy-Moderate. Wi-Fi to ThingSpeak/Blynk takes ~1 hour. Needs stable internet (common in India).
  • ML: Moderate (see below). Libraries simplify tasks.
  • Fabrication: Moderate. Sewing sensors requires care to avoid shorts. Waterproof with silicone sealant.
• Time:
  • Hardware: 1 week.
  • Software/ML: 1-2 weeks.
  • Testing/Integration: 1 week.
• Challenges:
  • Battery life (use sleep modes for 8-12 hours).
  • Sensor noise (e.g., motion affects HR; use filters).
  • Comfort (avoid bulky components).
  • Data privacy (encrypt with MQTT TLS).
• Beginner Tips: Follow YouTube tutorials (e.g., Random Nerd Tutorials). Test sensors individually. Prototype on breadboard to avoid sewing errors.

ML Knowledge Needed

• Level: Basic to Intermediate; no heavy ML required.
• Role: Detect anomalies (e.g., HR > 100 bpm, temp > 38°C). Run on edge (TensorFlow Lite on ESP32) for real-time or cloud (Google Colab) for complex analysis.
• Requirements:
  • Basic: Filter noise (moving averages). Use thresholds (e.g., if HR > 100, alert).
  • Intermediate: Train simple models (K-Means for clustering, Random Forest for risk classification) on datasets like MIT-BIH (ECG, free). Use TensorFlow Lite (edge) or scikit-learn (cloud). ~100-200 lines of code.
  • No Deep Learning: Pre-trained models give 90-95% accuracy without complex neural nets.
• Beginner Tips:
  • Start with thresholds (no ML).
  • Use Google Colab (free) for ML training.
  • Download pre-trained models from GitHub (e.g., HR anomaly detection).
• Resources: GitHub repos, TensorFlow Lite tutorials, YouTube (e.g., “ESP32 health ML”).

Getting Started (Beginner IoT Guide)

1. Learn Basics: Watch “ESP32 Arduino Tutorial” on YouTube (e.g., Random Nerd Tutorials). Understand I2C (for MAX30102, MPU-6050) and one-wire (DS18B20).
2. Prototype: Build on breadboard. Test sensors one-by-one using Arduino IDE sample code.
3. IoT Setup: Install Blynk app (free). Create gauges for HR/temp. Connect ESP32 to Wi-Fi (tutorials on Blynk.io).
4. ML: Use thresholds initially (e.g., Arduino: if (heartRate > 100) { alert(); }). Later, train a Random Forest model on Colab with MIT-BIH data. Deploy to ESP32 via TensorFlow Lite.
5. Jacket Integration: Sew sensors with conductive thread (Adafruit tutorials). Test comfort (wear for 1 hour). Waterproof electronics.
6. Testing: Wear jacket, verify data on Blynk, calibrate sensors (e.g., HR ~60-100 bpm).

Common Pitfalls

• Wiring Errors: Double-check with multimeter (borrow from lab).
• Sensor Noise: Use software filters (Arduino libraries).
• Battery Drain: Add sleep mode (ESP32 examples online).
• Wi-Fi Issues: Test with strong signal.

Sample Code

Find sample ESP32 code in smart_jacket.ino (reads sensors, sends to Blynk). Available in project repository or tutorials.

Resources

• Tutorials: Hackster.io, Instructables, YouTube (search “ESP32 health monitor”).
• Code: GitHub (e.g., “ESP32 health sensor”).
• Datasets: MIT-BIH ECG dataset (PhysioNet).
• Supplies: Amazon.in, Robu.in, local markets (Lamington Road, SP Road).